@evantahler/mcpx 0.15.0

package/.claude/settings.local.json

@@ -0,0 +1,18 @@
+{
+  "permissions": {
+    "allow": [
+      "WebFetch(domain:github.com)",
+      "Bash(bun lint:*)",
+      "Bash(bun:*)",
+      "WebFetch(domain:docs.arcade.dev)",
+      "Bash(grep:*)",
+      "WebSearch",
+      "Bash(wc:*)",
+      "Bash(gh release:*)",
+      "Bash(gh run:*)",
+      "Bash(mcpcli search:*)",
+      "Bash(mcpcli call:*)",
+      "Bash(mcpcli:*)"
+    ]
+  }
+}

package/.claude/skills/mcpx.md

@@ -0,0 +1,165 @@
+---
+name: mcpx
+description: Discover and use MCP tools via the mcpx CLI
+trigger: when the user wants to interact with external services, APIs, or MCP tools
+---
+
+# mcpx — MCP Tool Discovery and Execution
+
+You have access to external tools via `mcpx`. Use this workflow:
+
+## 1. Search for tools
+
+```bash
+mcpx search "<what you want to do>"
+```
+
+## 2. Inspect the tool schema
+
+```bash
+mcpx info <server> <tool>
+```
+
+This shows parameters, types, required fields, and the full JSON Schema.
+
+## 3. Execute the tool
+
+```bash
+mcpx exec <server> <tool> '<json args>'
+mcpx exec <server> <tool> -f params.json
+```
+
+## Rules
+
+- Always search before executing — don't assume tool names exist
+- Always inspect the schema before executing — validate you have the right arguments
+- Use `mcpx search -k` for exact name matching
+- Pipe results through `jq` when you need to extract specific fields
+- Use `-v` for verbose debugging (HTTP details + JSON-RPC protocol messages) if an exec fails unexpectedly
+- Use `-l debug` to see all server log messages, or `-l error` for errors only
+
+## Examples
+
+```bash
+# Find tools related to sending messages
+mcpx search "send a message"
+
+# See what parameters Slack_SendMessage needs
+mcpx info arcade Slack_SendMessage
+
+# Send a message
+mcpx exec arcade Slack_SendMessage '{"channel":"#general","message":"hello"}'
+
+# Chain commands — search repos and read the first result
+mcpx exec github search_repositories '{"query":"mcp"}' \
+  | jq -r '.content[0].text | fromjson | .items[0].full_name' \
+  | xargs -I {} mcpx exec github get_file_contents '{"owner":"{}","path":"README.md"}'
+
+# Read args from stdin
+echo '{"path":"./README.md"}' | mcpx exec filesystem read_file
+
+# Pipe from a file
+cat params.json | mcpx exec server tool
+
+# Read args from a file with --file flag
+mcpx exec filesystem read_file -f params.json
+```
+
+## 4. Long-running tools (Tasks)
+
+Some tools support async execution via MCP Tasks. mcpx auto-detects this and uses task-augmented execution when available.
+
+```bash
+# Default: waits for the task to complete, showing progress
+mcpx exec my-server long_running_tool '{"input": "data"}'
+
+# Return immediately with a task handle (for scripting/polling)
+mcpx exec my-server long_running_tool '{"input": "data"}' --no-wait
+
+# Check task status
+mcpx task get my-server <taskId>
+
+# Retrieve the result once complete
+mcpx task result my-server <taskId>
+
+# List all tasks on a server
+mcpx task list my-server
+
+# Cancel a running task
+mcpx task cancel my-server <taskId>
+```
+
+For tools that don't support tasks, `exec` works exactly as before.
+
+## 5. Elicitation (Server-Requested Input)
+
+Some servers request user input mid-operation (e.g., confirmations, auth flows). mcpx handles this automatically:
+
+```bash
+# Interactive — prompts appear in the terminal
+mcpx exec my-server deploy_tool '{"target": "staging"}'
+# Server requests input: Confirm deployment
+#   *Confirm [y/n]: y
+
+# Non-interactive — decline all elicitation (for scripts/CI)
+mcpx exec my-server deploy_tool '{"target": "staging"}' --no-interactive
+
+# JSON mode — read/write elicitation as JSON via stdin/stdout
+echo '{"action":"accept","content":{"confirm":true}}' | \
+  mcpx exec my-server deploy_tool '{"target": "staging"}' --json
+```
+
+## Authentication
+
+Some HTTP servers require OAuth. If you see an "Not authenticated" error:
+
+```bash
+mcpx auth <server>        # authenticate via browser
+mcpx auth <server> -s     # check token status and TTL
+mcpx auth <server> -r     # force token refresh
+mcpx deauth <server>      # remove stored auth
+```
+
+## Available commands
+
+| Command                                | Purpose                           |
+| -------------------------------------- | --------------------------------- |
+| `mcpx`                               | List all servers and tools        |
+| `mcpx servers`                       | List servers (name, type, detail) |
+| `mcpx -d`                            | List with descriptions            |
+| `mcpx info <server>`                 | Server overview (version, capabilities, tools) |
+| `mcpx info <server> <tool>`          | Show tool schema                  |
+| `mcpx exec <server>`                 | List tools for a server           |
+| `mcpx exec <server> <tool> '<json>'` | Execute a tool                    |
+| `mcpx exec <server> <tool> -f file`  | Execute with args from file       |
+| `mcpx search "<query>"`              | Search tools (keyword + semantic) |
+| `mcpx search -k "<pattern>"`         | Keyword/glob search only          |
+| `mcpx search -q "<query>"`           | Semantic search only              |
+| `mcpx search -n <number> "<query>"`  | Limit number of results (default: 10) |
+| `mcpx index`                         | Build/rebuild search index        |
+| `mcpx index -i`                      | Show index status                 |
+| `mcpx auth <server>`                 | Authenticate with OAuth           |
+| `mcpx auth <server> -s`              | Check token status and TTL        |
+| `mcpx auth <server> -r`              | Force token refresh               |
+| `mcpx deauth <server>`               | Remove stored authentication      |
+| `mcpx ping`                          | Check connectivity to all servers |
+| `mcpx ping <server> [server2...]`    | Check specific server(s)          |
+| `mcpx add <name> --command <cmd>`    | Add a stdio MCP server            |
+| `mcpx add <name> --url <url>`        | Add an HTTP MCP server            |
+| `mcpx add <name> --url <url> --transport sse` | Add a legacy SSE server  |
+| `mcpx remove <name>`                 | Remove an MCP server              |
+| `mcpx skill install --claude`        | Install mcpx skill for Claude   |
+| `mcpx skill install --cursor`        | Install mcpx rule for Cursor    |
+| `mcpx resource`                     | List all resources across servers |
+| `mcpx resource <server>`            | List resources for a server       |
+| `mcpx resource <server> <uri>`      | Read a specific resource          |
+| `mcpx prompt`                       | List all prompts across servers   |
+| `mcpx prompt <server>`              | List prompts for a server         |
+| `mcpx prompt <server> <name> '<json>'` | Get a specific prompt          |
+| `mcpx exec <server> <tool> --no-wait` | Execute as async task, return handle |
+| `mcpx exec <server> <tool> --ttl <ms>` | Set task TTL (default: 60000) |
+| `mcpx -N exec <server> <tool> ...`  | Decline elicitation (non-interactive) |
+| `mcpx task list <server>`            | List tasks on a server          |
+| `mcpx task get <server> <taskId>`    | Get task status                 |
+| `mcpx task result <server> <taskId>` | Retrieve completed task result  |
+| `mcpx task cancel <server> <taskId>` | Cancel a running task           |

package/.claude/worktrees/elastic-jennings/.claude/settings.local.json

@@ -0,0 +1,18 @@
+{
+  "permissions": {
+    "allow": [
+      "WebFetch(domain:github.com)",
+      "Bash(bun lint:*)",
+      "Bash(bun:*)",
+      "WebFetch(domain:docs.arcade.dev)",
+      "Bash(grep:*)",
+      "WebSearch",
+      "Bash(wc:*)",
+      "Bash(gh release:*)",
+      "Bash(gh run:*)",
+      "Bash(mcpcli search:*)",
+      "Bash(mcpcli call:*)",
+      "Bash(mcpcli:*)"
+    ]
+  }
+}

package/.claude/worktrees/elastic-jennings/.claude/skills/mcpcli.md

@@ -0,0 +1,93 @@
+---
+name: mcpcli
+description: Discover and use MCP tools via the mcpcli CLI
+trigger: when the user wants to interact with external services, APIs, or MCP tools
+---
+
+# mcpcli — MCP Tool Discovery and Execution
+
+You have access to external tools via `mcpcli`. Use this workflow:
+
+## 1. Search for tools
+
+```bash
+mcpcli search "<what you want to do>"
+```
+
+## 2. Inspect the tool schema
+
+```bash
+mcpcli info <server> <tool>
+```
+
+This shows parameters, types, required fields, and the full JSON Schema.
+
+## 3. Execute the tool
+
+```bash
+mcpcli exec <server> <tool> '<json args>'
+```
+
+## Rules
+
+- Always search before executing — don't assume tool names exist
+- Always inspect the schema before executing — validate you have the right arguments
+- Use `mcpcli search -k` for exact name matching
+- Pipe results through `jq` when you need to extract specific fields
+- Use `-v` for verbose HTTP debugging if an exec fails unexpectedly
+
+## Examples
+
+```bash
+# Find tools related to sending messages
+mcpcli search "send a message"
+
+# See what parameters Slack_SendMessage needs
+mcpcli info arcade Slack_SendMessage
+
+# Send a message
+mcpcli exec arcade Slack_SendMessage '{"channel":"#general","message":"hello"}'
+
+# Chain commands — search repos and read the first result
+mcpcli exec github search_repositories '{"query":"mcp"}' \
+  | jq -r '.content[0].text | fromjson | .items[0].full_name' \
+  | xargs -I {} mcpcli exec github get_file_contents '{"owner":"{}","path":"README.md"}'
+
+# Read args from stdin
+echo '{"path":"./README.md"}' | mcpcli exec filesystem read_file
+```
+
+## Authentication
+
+Some HTTP servers require OAuth. If you see an "Not authenticated" error:
+
+```bash
+mcpcli auth <server>        # authenticate via browser
+mcpcli auth <server> -s     # check token status and TTL
+mcpcli auth <server> -r     # force token refresh
+mcpcli deauth <server>      # remove stored auth
+```
+
+## Available commands
+
+| Command                                | Purpose                           |
+| -------------------------------------- | --------------------------------- |
+| `mcpcli`                               | List all servers and tools        |
+| `mcpcli -d`                            | List with descriptions            |
+| `mcpcli info <server>`                 | Show tools for a server           |
+| `mcpcli info <server> <tool>`          | Show tool schema                  |
+| `mcpcli exec <server>`                 | List tools for a server           |
+| `mcpcli exec <server> <tool> '<json>'` | Execute a tool                    |
+| `mcpcli search "<query>"`              | Search tools (keyword + semantic) |
+| `mcpcli search -k "<pattern>"`         | Keyword/glob search only          |
+| `mcpcli search -q "<query>"`           | Semantic search only              |
+| `mcpcli index`                         | Build/rebuild search index        |
+| `mcpcli index -i`                      | Show index status                 |
+| `mcpcli auth <server>`                 | Authenticate with OAuth           |
+| `mcpcli auth <server> -s`              | Check token status and TTL        |
+| `mcpcli auth <server> -r`              | Force token refresh               |
+| `mcpcli deauth <server>`               | Remove stored authentication      |
+| `mcpcli add <name> --command <cmd>`    | Add a stdio MCP server            |
+| `mcpcli add <name> --url <url>`        | Add an HTTP MCP server            |
+| `mcpcli remove <name>`                 | Remove an MCP server              |
+| `mcpcli skill install --claude`        | Install mcpcli skill for Claude   |

package/.claude/worktrees/elastic-jennings/.github/workflows/auto-release.yml

@@ -0,0 +1,117 @@
+name: Auto Release
+
+on:
+  push:
+    branches: [main]
+  workflow_dispatch:
+
+permissions:
+  contents: write
+  id-token: write
+
+jobs:
+  check-version:
+    runs-on: ubuntu-latest
+    outputs:
+      should_release: ${{ steps.check.outputs.should_release }}
+      tag: ${{ steps.version.outputs.tag }}
+      version: ${{ steps.version.outputs.version }}
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Get version from package.json
+        id: version
+        run: |
+          VERSION=$(jq -r .version package.json)
+          echo "version=$VERSION" >> "$GITHUB_OUTPUT"
+          echo "tag=v$VERSION" >> "$GITHUB_OUTPUT"
+
+      - name: Check if release exists
+        id: check
+        run: |
+          if gh release view "${{ steps.version.outputs.tag }}" &>/dev/null; then
+            echo "should_release=false" >> "$GITHUB_OUTPUT"
+          else
+            echo "should_release=true" >> "$GITHUB_OUTPUT"
+          fi
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+  create-release:
+    needs: check-version
+    if: needs.check-version.outputs.should_release == 'true'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Create release
+        run: |
+          gh release create "${{ needs.check-version.outputs.tag }}" \
+            --title "${{ needs.check-version.outputs.tag }}" \
+            --generate-notes
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+  ci:
+    needs: [check-version, create-release]
+    if: needs.check-version.outputs.should_release == 'true'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+      - run: bun install --frozen-lockfile
+      - run: bun lint
+      - run: bun test
+
+  publish-npm:
+    needs: [check-version, ci]
+    if: needs.check-version.outputs.should_release == 'true'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      id-token: write
+    steps:
+      - uses: actions/checkout@v4
+      - uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 22
+          registry-url: https://registry.npmjs.org
+      - run: npm install -g npm@latest
+      - run: bun install --frozen-lockfile
+      - run: npm publish --provenance --access public
+
+  build-binaries:
+    needs: [check-version, ci]
+    if: needs.check-version.outputs.should_release == 'true'
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        include:
+          - target: bun-darwin-arm64
+            artifact: mcpcli-darwin-arm64
+          - target: bun-darwin-x64
+            artifact: mcpcli-darwin-x64
+          - target: bun-linux-arm64
+            artifact: mcpcli-linux-arm64
+          - target: bun-linux-x64
+            artifact: mcpcli-linux-x64
+    steps:
+      - uses: actions/checkout@v4
+      - uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+      - run: bun install --frozen-lockfile
+      - name: Build binary
+        run: |
+          bun build --compile --minify --sourcemap \
+            --target=${{ matrix.target }} \
+            --define BUILD_VERSION="'\"${{ needs.check-version.outputs.version }}\"'" \
+            ./src/cli.ts --outfile dist/${{ matrix.artifact }}
+      - name: Upload to release
+        run: gh release upload "${{ needs.check-version.outputs.tag }}" dist/${{ matrix.artifact }} --clobber
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}

package/.claude/worktrees/elastic-jennings/.github/workflows/ci.yml

@@ -0,0 +1,18 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+jobs:
+  check:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+      - run: bun install --frozen-lockfile
+      - run: bun lint
+      - run: bun test

package/.claude/worktrees/elastic-jennings/.prettierignore

@@ -0,0 +1,4 @@
+node_modules
+dist
+bun.lock
+.claude

package/.claude/worktrees/elastic-jennings/.prettierrc

@@ -0,0 +1,7 @@
+{
+  "semi": true,
+  "singleQuote": false,
+  "trailingComma": "all",
+  "printWidth": 100,
+  "tabWidth": 2
+}

package/.claude/worktrees/elastic-jennings/CLAUDE.md

@@ -0,0 +1,19 @@
+# CLAUDE.md
+
+## Project
+
+mcpcli — A CLI for MCP servers. "curl for MCP."
+
+## Commands
+
+- `bun run dev` — Run in development mode
+- `bun test` — Run tests
+- `bun lint` — Check formatting (prettier)
+- `bun format` — Auto-fix formatting
+- `bun run build` — Build single binary
+
+## Rules
+
+- **Always bump the patch version in `package.json`** when making any code changes (source, tests, config). Use semver: patch for fixes/small changes, minor for new features, major for breaking changes.
+- **Always keep `README.md` and `.claude/skills/mcpcli.md` in sync** with any CLI changes (commands, flags, syntax, examples). The skill file includes workflow steps, code examples, and a command table that must all reflect the current CLI surface.
+- **Always run `bun run format`** before committing to fix prettier formatting issues.

package/.claude/worktrees/elastic-jennings/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Evan Tahler
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.